Pt Oxygen example

This is an example of a simple irreversible reaction where oxygen is fed over a platinum catalyst. The goal is to show the amount of conversion of the oxygen in a robust manner. The problem with the dataset is that it includes many outliers and the mass spectrometer measurements has a significant drift from pulse to pulse. This example will use TEAK to determine the appropriate conversion values. Later in the example, the traditional calibration methods will be used and compare to TEAK.

Calibration via TEAK

The inert gas in this experiment is Argon and the reactant is Oxygen. To perform TEAK, the Argon information will first be baseline corrected using information about the residence time distribution and then calibrated to a single reference flux.

Just looking at the Argon M0, outgassing can be observed at several pulse indices.

Plots of the outgassing examples

Even after removing the outliers from the data, there still could be room for improvement shown in th eplot below. However, since the M0 values are relatively close (0.53 to 0.533), no further pruning of the data will be performed.

As a robust measure of the variability, the Median Absolute Deviation (MAD) is given below showing minimal variation pulse to pulse.

After the Argon data is calibrated, the oxygen data will be baseline corrected and then calibrated to the Argon information. A huber_loss is set to True to account for any significant outliers within the individual flux.

The calibrated oxygen shows potential outgassing as distinct spikes in the M0 over the pulse index.

Below is subset of the M0 spikes plotted as flux over time. Outliers in the lower in the pulse index can be attributed to low signal to noise ratios while the larger pulse indices clearly exhibit outgassing.

These outliers can easily be deteced by looking at where the difference of the M0 is significantly larger than the standard deviation of the differenced M0. Below is a plot of the M0 values where the outliers are removed.

With the TEAK preprocessing, the flux comparison below shows that by the end of the experiment, only transport is occuring with the oxygen.

Calibration via Traditional baseline correction and calibration

The same data as above will be calibrated via the traditional methodology of baseline correction (using a time range) and calibration (using information about an inert experiment). Note that the traditional method is much faster than TEAK in computation time.

Baseline correcting both Argon and Oxygen from 3.8 to 4 seconds.

Applying the inert experiment calibration value (1 / 0.761) to the calibration of the Oxygen. Just as a reminder, an inert experiment calibration value is determined as the ratio of the means of Argon and Oxygen over a series of pulses without a catalyst present.

Plotting the Argon and the Oxygen moments while removing the same outgassing pulses determine by TEAK.

Below is a plot of the last pulse within the traditional calibration. Note that even though the flux match, the traditional calibration underestimates the conversion.

Comparison of the conversion values

Plotting the mean residence time comparison.

Plotting the variance residence time (the variance should never be negative). Any negative value of the variance will lead to miscalculation of the residence time coefficients as the formula uses the M0, M1 and M2.

Futher focusing on where the conversion is less than 90%.